Painting the near surface using geology, geophysics, and satellites

Abstract

A smooth interpolation of a sparse uphole survey may be the main tool available for modeling the near surface in desert areas. An uphole survey does not properly define sharp, near-surface velocity changes associated with lithologic and topographic variations such as cliffs, canyons, or outcrops, which may be delineated by satellite imagery instead. We present a patchy interpolation technique to build a consistent near-surface model. The algorithm is based on [Formula: see text]-means clustering to integrate geological, geophysical, and remote-sensing information. A somewhat arbitrary aspect of the method is the choice of how many clusters are used in segregating the data. Linking this parameter to surface geologic formations may be inadequate because P- or S-wave velocities are unlikely to have a sufficiently unique correlation with geologic age and lithology. We suggest an empirical criterion: improvement in the seismic stack section achieved by processing with parameters derived by various clustering choices and data-type combinations. This approach was tested using field data from Saudi Arabia.